神经网络特征属性反演在煤系地层致密储层预测中的应用——以鄂尔多斯盆地东缘神府木瓜区为例  

Application of neural network feature attribute inversion in the prediction of tight reservoirs in coal-bearing strata:A case study in the Shenfu Mugua area on the eastern margin of the Ordos Basin

作　　者：史浩 刘小波[2] 杨桂茹[3] 林利明 刘玉梅[3] 张发强[2] 王振国 郭俊超 SHI Hao;LIU Xiaobo;YANG Guiru;LIN Liming;LIU Yumei;ZHANG Faqiang;WANG Zhenguo;GUO Junchao(Zhonglian Coalbed Methane Company Limited,Beijing 100015 China;Northeast Petroleum University,Daqing 163318 China;PetroChina Research Institute of Petroleum Exploration and Development,Beijing 100083 China)

机构地区：[1]中联煤层气有限责任公司,北京100015 [2]东北石油大学,黑龙江大庆163318 [3]中国石油勘探开发研究院,北京100083

出　　处：《天然气地球科学》2024年第10期1886-1896,共11页Natural Gas Geoscience

基　　金：中海石油中国有限公司重大科技专项“中联公司上产60亿方关键技术研究”(编号:CNOOC-KJ135ZDXM40);中联煤层气有限责任公司自主立项科研项目“神府区块木瓜区煤系地层致密气储层高精度识别及关键技术研究”联合资助。

摘　　要：煤系地层致密储层叠后地震预测技术备受关注。由于致密储层的地震分辨率低、地球物理特征微弱导致其难以识别等特点,因此,煤系地层致密储层地震预测的精确性较低。以鄂尔多斯盆地东缘神府木瓜区太原组煤系地层为研究对象,利用神经网络的多源数据融合分析学习能力,采用叠后波阻抗反演及叠前弹性参数反演等多源数据融合消除煤层干扰,建立了基于神经网络特征属性的煤系地层致密储层逐级预测方法。该方法对木瓜区太原组煤系地层致密储层弹性参数、孔隙度和含气量等进行了成功预测,验证井符合度均达到80%以上,能充分利用各类型地震资料的敏感优势,对储层多种地质属性进行感知预测,以实现煤系地层致密储层的精准刻画,为致密储层天然气勘探开发提供技术支持。The post-stack seismic prediction technology of coal-bearing strata reservoirs has attracted wide⁃spread attention.However,due to the low seismic resolution and the difficulty in identifying weak geophysical features in tight reservoirs,the accuracy of seismic prediction of coal-bearing strata reservoirs is relatively low.This study focuses on the Taiyuan Formation coal-bearing strata in the Shengfu Mugua area on the eastern mar⁃gin of the Ordos Basin.Utilizing the multi-source data fusion analysis and learning capabilities of neural net⁃works,and employing the fusion of post-stack wave impedance inversion and pre-stack elastic parameter inver⁃sion to eliminate coal seam interference,a step-by-step prediction method for tight reservoirs in coal-bearing strata based on neural network feature attributes has been established.This method successfully predicted the elastic parameters,porosity,and gas content of the tight reservoirs in the Taiyuan Formation coal-bearing strata in the Mugua area,with well logging matching rates all exceeding 80%.The method can fully leverage the sen⁃sitive advantages of various types of seismic data to perceptively predict multiple geological properties of reser⁃voirs,achieving accurate characterization of tight reservoirs in coal-bearing strata.This approach can provide technical support for the exploration and development of natural gas in tight reservoirs in China.

关 键 词：神经网络 属性反演 天然气 致密气 致密砂岩 煤层 储层预测 

分 类 号：TE121.1[石油与天然气工程—油气勘探]